I assure you, it’s not lack of ability.
2 Likes
I hate to sound like a broken record, but people in the real world are no different than they are in OW. They belong to the same set: “people.” And so solutions that work in the real world WILL work in OW (whether the real world has the will to do anything about a particularly problematic group of people is a different story).
But anyway, the first you thing you do is you tag the smurf. Anyone exhibiting smurf-like behavior now has a marker on their account, which gives way to a probationary period. They are not informed about this period through some warning or whatnot. You monitor what they do. Get to plat and hard throw their way back down to silver? Mark it. Get to plat and soft throw their way down to silver? Mark it. Smurfs are not going to be playing like a normal player at that rank who’s actually trying: the data is just not going to look to the same, and if they ARE playing like a typical player at that rank, we no longer have a smurf problem because their behavior is no longer problematically anomalous.
During this period you look for wild swings in behavior (obvious throws, not so obvious throws, tilt-driven popping off, whatever the data tells you is “smurf-like”). During the probationary period, you acquire more information which increases or decreases your confidence about this person beyond some threshold.
Then it becomes about research and A/B split testing. You know that you want to de-incentivize this behavior but you may not know how. OK, fine. You try different things and see what works. The things you mentioned are all fine. I don’t think this is an “aha!” moment where someone stands up during a Bliz meeting and shouts “Eureka! I’ve found the answer to smurfing!” I think you try a bunch of things until you find what has a chilling effect on this behavior. I mean logically, once you’ve found a smurf, the question becomes “is there ANYTHING we can do to make this person’s OW life more difficult and unpleasant, up to and including progressively timing-out and eventually banning the account.” (Substantially) longer queue times. Tougher lobbies. “Smurf, leavers, afk, high ping, toxic lobbies.” Phone verification. All options are on the table. We make prison suck so much that people don’t want to be there, the same principle applies here.
We agreed that FINDING smurfs would not be the limiting factor. And we know logically that we have up to and including banning the account at our disposal. Doing anything in between which makes this person’s experience awful should be considered until you find a methodology that works.
2 Likes
You can do all that complicated stuff…or…drumroll…what I said above. Two-factor authentication required to enter Ranked queue. VERY FEW people are going to carry $30/mo. phone accounts just so they can smash lower-ranked players.
2 Likes
I had a few wins. My losses mostly seemed like bad team synergy more than anything though. I had two games with healers raging in placements which seemed to cause problems and throw off the team and ended up making us lose but eh. 
Sure, I get that. What I was trying to say is that what I usually like to do when thinking about a particular issue is set a premise (smurfs are a problem) and then try to come up with a personally ideal solution on how to solve the problem. Once I have done that, I would compare my solution to the real world implementation. Often I would then realize that I my solution was missing something or that a component that I had been thinking about interacted with a different one causing unwanted side effects.
I thought it was interesting you would suggest something like marking or flagging accounts. This is essentially what some people already think exists in the game, and argue against. To be honest, I don’t think we would need to do this. As you noted, a smurf will have a wide range of a-typcial stats and any matchmaking developer should be able to account for these stats by means of PBSR for instance. I’ll elaborate on this below.
Here, I wanna ask if you really wanna make it more unpleasant for the smurf? Ideally, you just wanna move them to their proper rank as fast as possible. This way, you don’t piss them off (since they are a customer willing to spend extra money) and you don’t piss off the rest of the community who are not smurfing. A possible implementation of this could then be PBSR by assigning much more SR gains for wins, and when the player throws to lose SR, you maintain their MMR. This way, the smurf’s SR difference for W/Ls will get greater with every ‘throwing cycle’ until they would lose 1-10 SR for a loss and 40-50 SR for a win.
What I just described is my ideal system that would kick in after apparent a-typical behavior of a particular account after a sufficient amount of games. If we compare this to OW’s current implementation, I think we will notice that PBSR has been introduced for that same purpose although in a tamer form. This, then, raises the question why is it tamer? Maybe they don’t wanna commit to an extreme model like mine because they can’t detect a smurf with certainty after all (note that a good smurf can throw very subtly: not finishing that big value stagger kill 10 seconds before the game ends, intentionally ignoring a high priority opponent to force a loss, etc.). Or maybe there is a huge unwanted interaction of my proposal with something else.
Banning an account is the last thing I would do, both, in my ideal system as well as in a real world implementation. You don’t wanna punish people wanting to play your game. Smurfing can also have all sorts of reasons, like wanting to play with friends that you either freshly introduced to the game or are just lower ranked than you. I know you can make an argument for these people too, but I think it’s a mistake shoving them into the same category as, for instance, cheaters.
So in the end, I think we have a somewhat similar methodology; though, your goal is to make the experience of smurfs as unpleasant as possible, whereas mine would be to have the smurfs succeed at playing to the degree where it just doesn’t make sense to not play on their main.
5 Likes
Don’t you have something like 20 smurf/alt accounts?
Naturally you’d be more lenient towards them.
There’s nothing wrong with people having multiple accounts. They just shouldn’t be allowed in lower-level ranked play.
1 Like
I do, but that’s irrelevant to the discussion.
3 Likes
Conflict of interest. Should be stated for full disclosure.
1 Like
There isn’t a conflict of interest. I’d give up my accounts in an instant if that meant a healthier competitive ladder.
3 Likes
You can already do that now can you not?
1 Like
Its like you are implying Basil is the only one with multiple accounts, and it will suddenly solve the issue.
Also funny its coming from you, the guy with 2-3 accounts all hardstruck low diamond and below
2 Likes
The games I play below my main total to <10 games per season. I could stop playing those games, too; though, I don’t see any reason to do that considering ow1 will stop existing soon and the few players that are left playing the game don’t care anymore anyway.
2 Likes
Non-factual speculation.
1 Like
it’s speculation but I would argue that the chance of it being factual is bigger the chance of it not being factual.
3 Likes
What data or evidence do you base this hypothesis on?
1 Like
Makes sense to me. If we get down to first principles, this problem has two dimensions: 1) properly identifying the smurf, and 2) disincentivizing the behavior. We agree that 1) isn’t rocket science. In order to address 2) we consider what the smurfs wants and then prohibit them from getting it.
This is much easier if we have high levels of confidence in step 1) because then we can be assured that we aren’t making some random person’s in-game life a living hell. I actually think a message from Blizzard along the lines of “your account has been flagged for unusual behavior” at some point in this timeline, would be a fairly unsettling message, and is most likely to going to prompt SOME behavior adjustment from the person receiving it. No matter how much a person doesn’t care about an account, leveling up a new account or buying a new one isn’t something anyone wants to do over and over; particularly, if they know they’re just going to run into the same roadblock with that new account. What would be the the point of acquiring a new one?
If the person DOES actually modify their behavior based on some ominous Blizzard warning, ironically, we have even MORE confidence now that this person is a smurf. We can then decrease the time between more warnings and lengthen the punishment. And then if it keeps happening, of course they just get a ban. IP blocks. Phone verification. Hardware bans. Etc.
Hmm, tagging the person has no downside, but not tagging them does. Not tagging them throws off the confidence level and disconnects the person from their history (or it just creates more computing overhead as everyone’s history has to be repeatedly re-queried for anomalous behavior). This is why convicts have prison records, HR people have files on under-performing employees,etc. it links behavior with individuals, and lowers the chances that they become unlinked from past behavior, lowering false positives and false negatives through having to re-identify them from scratch.
There are a lot of bad arguments from the “rigged conspirators.” 
And even I don’t think the game is “rigged” in the typical sense: “let’s make this guy or this set of people lose.” I think bad matchmaking/entropy is a form of grind/engagement. But again, I don’t think the matchmaking is intentionally bad, such that devs are sitting around making it deliberately worse.
I think devs look at play time, they look at churn, and come to some conclusions about how long the average person will pursue some goal before they give up. This is the exact same thing pretty much ALL developers do, but the way they determine this has a lot to do with the game’s format. Take some boss-based game like Elden Ring with an extensive progression system, etc. Devs are going to tune that end-level boss based on how long it takes the typical person to progress past it. That could be tuned for 30 mins or for 3 hours. If they set it to 5 hours and their internal play testers hate it, they’ll adjust it down to say 4. If the boss is particularly fun and exciting, hell, play testers may love struggling against this boss for 6 hours or more (including re-exploring the world to level up their character). I think the exact same kind thing is happening in OW. But people in OW were willing to grind way longer than anyone at Blizzard imagined. Meaning that wherever the difficulty was set to became de facto “favorable” even if players hated it (yet kept playing). This is my theory as to why matches are inconsistent and climbing is so hard. On the whole, climbing difficulty is not set based on what’s fair or competitive, but based on RESULTS and ENGAGEMENT.
I consider myself a reasonable person, but in retrospect, I see the number of hours I put into climbing so idiotic that I’m now like “who the hell was that guy?” This is telling to me. We have agreed countless time that skill matters. But let’s assume for a second, just humor me here, that the degree of difficulty involved in getting out of gold WAS too high, but players were so committed to achieving this result that they just kept grinding, driving engagement, what does the dev team do? Engagement is up. People are in the practice range pew-pew-ing away. I think Blizzard says “Eureka, we’ve found our balance” and that’s that. I really don’t know that the opinion of a GM progressing through gold really tells us much. A while back I’d mentioned that if you can only lift 100 lbs, 105 will seem impossible. If you can lift 300lbs, you will not even notice the difference between 100 and 105. To some extent, I think there’s a fair bit of this phenomenon happening when high ranked players assess what it’s like to play far below their own rank.
I think this comes down to a subjective assessment of how detrimental smurfs are, what their objective impact is, what their perceived impact (within the community) is, and then more objective things like, given their relatively small numbers (but their ability to affect 11 other players per match, further distorting rank accuracy) is it worth tolerating them at all, whether they conceptually have a proper, justifiable existence anywhere in the game, etc.
I like your suggestion and am totally all for Blizzard TRYING different things to see what works. I think lawmakers have to do this all the time. Initial sentencing wasn’t having the desired effect, so they came up with the “three strikes” rule which made felons with two prior felonies seriously consider whether a third incident was worth spending the rest of their lives in prison. I think this process needs to be exploratory if nothing else. The idea proposed by some, “that there’s nothing that can be done about smurfs” just seems illogical. And in a sense, the smurfs themselves will let you know when you’ve gone far enough because their numbers will drop once you’ve gotten the methodology right.
I don’t think smurfs want to play this way: carefully playing like a plat, pretty much all the time, to avoid detection, and then finally, subtly throwing the match. I think this largely defeats the purpose of smurfing.
I think they like pissing people off, revenge-popping off, tilt-popping off, letting other players know they’re being screwed with one way or another, etc etc all of which are game-play signatures that are abrupt and obvious and “fun.” I honestly don’t think from an AI/Algorithmic point of view that there’s ANYTHING a smurf could do to throw off a well-designed system. From mouse sense, to looking around the map, to bursts of accuracy and tracking, to when they pop ults, to lack of feeding, to positioning, to habitual pragmatic reticle placement, etc etc. I’m a junior programmer and have started studying statistics. You put a team of professional, seasoned statisticians with software engineering backgrounds on the case, and equip them with with AI and Machine learning (ML) tools, and trust me, there’s nothing a smurf can do to hide their behavior. They simply are fundamentally not like the players they’re sandbagging against, and they’ll give that away in ways they couldn’t even imagine when investigated by AI/ML, things you and I (or they) would never even think to look for.
I hear you. I think the smurfs essentially teach Blizzard how to treat them and how to proceed. If they persist and persist and persist and connive and evade and trick in the face of Blizzard truly not wanting them around, I think they basically set the stage for how that “confrontation” with Blizzard plays out as well as the measures taken.
1 Like
Blizzard has to rework the matchmaking system. Its so unfair when every new account stomps our team and we lose every single time.
Watching smurfs try to defend their actions is as comical as it is shameful.
You also have multiple accounts.
Let’s not throw stones from glass houses
2 Likes